摘要
针对瓦斯突出等级评判方法预测准确度低的问题,提出一种基于拉普拉斯特征映射算法(LE)和改进的乌鸦搜索算法(ICSA)优化核极限学习机(KELM)的瓦斯突出预测模型。利用LE算法对瓦斯突出数据进行非线性降维,消除变量间的相互重叠;引入Tent扰动序列、自适应步长和自适应感知概率改进传统的乌鸦搜索算法(CSA),有效避免算法陷入局部最优,提高算法的收敛性能;采用ICSA算法对KELM的相关参数进行寻优,建立基于LE和ICSA-KELM的瓦斯突出等级评判模型。经过对比试验表明,该模型能够有效提高预测准确率。
Aiming at the problem of low prediction accuracy of traditional gas outburst grade evaluation method,a gas outburst prediction model based on LE algorithm and ICSA algorithm optimized KELM was proposed.LE algorithm was used to reduce the dimension of gas outburst data nonlinearly and eliminate the overlap between variables;the traditional CSA algorithm was improved by adding Tent perturbation sequence,adaptive step size and adaptive perception probability.It can avoid failing into local optimum effectively,and improve the convergence performance of the algorithm.The ICSA algorithm was used to optimize the relevant parameters of KELM,and the grade evaluation model of gas outburst based on LE and ICSA-KELM was established.The experimental results show that this model can improve the prediction accuracy effectively.
作者
赵国强
王留洋
刘雨竹
卢万杰
王志中
ZHAO Guoqiang;WANG Liuyang;LIU Yuzhu;LU Wanjie;WANG Zhizhong(Faculty of Electrical and Control Engineering,Liaoning Technology University,Huludao 125105,China;School of Mechanical Engineering,Liaoning Technology University,Fuxin 123000,China)
出处
《辽宁工程技术大学学报（自然科学版）》
CAS
北大核心
2023年第1期32-39,共8页
Journal of Liaoning Technical University (Natural Science)
基金
国家自然科学基金(51974151,71771111)
辽宁省高等学校创新团队项目(LT2019007)
